—The main challenges in smart camera networks come from the limited capacity of network communications. Indeed, wireless protocols such as the IEEE 802.15.4 protocol target low data rate, low power consumption and low cost wireless networking in order to fit the requirements of sensor networks. Since nodes more and more often integrate image sensors, network bandwidth has become a strong limiting factor in application deployment. This means that data must be processed at the node level before being sent on the network. In this context, FPGA-based platforms, supporting massive data parallelism, offer large opportunities for on-board processing. We present in this paper our FPGA-based smart camera platform, called DreamCam, which is able to autonomously exchange processed information on an Ethernet network.